Biological networks:Types and origin

Protein-protein interactions, complexes, and network properties

Thomas Skøt JensenCenter for Biological Sequence AnalysisThe Technical University of Denmark

Networks in electronics

Radio kindly provided by Lazebnik, Cancer Cell, 2002

Model Generation

Interactions

Radio kindly provided by Lazebnik, Cancer Cell, 2002

Parts List

YER001WYBR088CYOL007CYPL127CYNR009WYDR224CYDL003WYBL003C…

YDR097CYBR089WYBR054WYMR215WYBR071WYBL002WYNL283CYGR152C…

• Sequencing

• Gene knock-out

• Microarrays

• etc.

Interactions

• Genetic interactions

• Protein-Protein interactions

• Protein-DNA interactions

• Subcellular Localization

Dynamics

• Microarrays

• Proteomics

• Metabolomics

Types of networks

Interaction networks in molecular biology

• Protein-protein interactions• Protein-DNA interactions• Genetic interactions• Metabolic reactions• Co-expression interactions• Text mining interactions• Association networks

Interaction networks in molecular biology

• Protein-protein interactions• Protein-DNA interactions• Genetic interactions• Metabolic reactions• Co-expression interactions• Text mining interactions• Association networks

Characterization of physical interactions

• Obligation– obligate (protomers only found/function together)– non-obligate (protomers can exist/function alone)

• Time of interaction– permanent (complexes, often obligate)– strong transient (require trigger, e.g. G proteins)– weak transient (dynamic equilibrium)

ol

Examples: GPCR

obligate, permanent

non-obligate,

strong transient

Approaches by interaction type

• Physical Interactions– Yeast two hybrid screens– Affinity purification (mass spec)– Protein-DNA by chIP-chip

• Other measures of ‘association’– Genetic interactions (double deletion

mutants)

– Functional associations (STRING)– Co-expression

Yeast two-hybrid method

Y2H assays interactions in vivo.

Uses property that transcription factors generally have separable transcriptional activation (AD) and DNA binding (DBD) domains.

A functional transcription factor can be created if a separately expressed AD can be made to interact with a DBD.

A protein ‘bait’ B is fused to a DBD and screened against a library of protein “preys”, each fused to a AD.

Yeast two-hybrid method

Fields and Song

Issues with Y2H

• Strengths– High sensitivity (transient & permanent PPIs)– Takes place in vivo– Independent of endogenous expression

• Weaknesses: False positive interactions– Auto-activation– ‘sticky’ prey– Detects “possible interactions” that may not take place under real

physiological conditions– May identify indirect interactions (A-C-B)

• Weaknesses: False negatives interactions– Similar studies often reveal very different sets of interacting proteins (i.e.

False negatives)– May miss PPIs that require other factors to be present (e.g. ligands,

proteins, PTMs)

Protein interactions by immuno-precipitation followed by mass spectrometry

• Start with affinity purification of a single epitope-tagged protein

• This enriched sample typically has a low enough complexity to be fractionated on a standard polyacrylamide gel

• Individual bands can be excised from the gel and identified with mass spectrometry.

Affinity Purification

Affinity PurificationStrengths

• High specificity

• Well suited for detecting permanent or strong transient interactions (complexes)

• Detects real, physiologically relevant PPIs

Weaknesses

• Less suited for detecting weaker transient interactions (low sensitivity)

• May miss complexes not present under the given experimental conditions (low sensitivity)

• May identify indirect interactions (A-C-B)

Protein-protein interaction data growth

Error rate may be as high as 30-50 %

Topology based scoring of interactions

Low confidence (4 unshared interaction partners)High confidence (1 unshared interaction partners)

A B C

Yeast two-hybrid

Low confidence (rarely purified together)

High confidence (often purified together)

Complex pull-downs

D

de Lichtenberg et al., Science, 2005

Filtering by subcellular localization

de Lichtenberg et al., Science, 2005

Filtering reduces coverage and increases specificity

Network Properties

Graphs, paths, topology

Graphs

• Graph G=(V,E) is a set of vertices V and edges E

• A subgraph G’ of G is induced by some V’ V and E’ E

• Graph properties:– Connectivity (node degree, paths)– Cyclic vs. acyclic– Directed vs. undirected

Sparse vs Dense

• G(V, E) where |V|=n, |E|=m the number of vertices and edges

• Graph is sparse if m~n

• Graph is dense if m~n2

• Complete graph when m=n2

Connected Components

• G(V,E)

• |V| = 69

• |E| = 71

Connected Components

• G(V,E)

• |V| = 69

• |E| = 71

• 6 connected components

Paths

A path is a sequence {x1, x2,…, xn} such that (x1,x2), (x2,x3), …, (xn-1,xn) are edges of the graph.

A closed path xn=x1 on a graph is called a graph cycle or circuit.

Shortest-Path between nodes

Shortest-Path between nodes

Longest Shortest-Path

Degree or connectivity

Random vs scale-free networks

P(k) is probability of each degree k, i.e fraction of nodes having that degree.

For random networks, P(k) is normally distributed.

For real networks the distribution is often a power-law:

P(k) ~ k

Such networks are said to be scale-free

“The Swedish sex web”

Target the ‘hubs’ to have an efficient safe sex education campaign

Lewin Bo, et al., Sex i Sverige; Om sexuallivet i Sverige 1996, Folkhälsoinstitutet, 1998

y = 1.2x-1.91

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1.0E+00

1.0E+01

1 10 100

Degree k

P (k

)Knock-out lethality and

connectivity

0

10

20

30

40

50

60

0 5 10 15 20 25

Degree k

% E

ssen

tial G

enes

Clustering coefficient

12

2

kk

nkn

C III

k: neighbors of I

nI: edges between

node I’s neighbors

The density of the network surrounding node I, characterized as the number of triangles through I. Related to network modularity

The center node has 8 (grey) neighbors

There are 4 edges between the neighbors

C = 2*4 /(8*(8-1)) = 8/56 = 1/7

Protein complexes have a high clustering coefficient

Proteins subunits are highly interconnected and thus have a high

clustering coefficient

There exists algorithms, such as MCODE, for identifying subnetworks (complexes) in large protein-protein

interaction networks

Hierarchical Networks

Detecting hierarchical organization

Scale-free networks are robust

• Complex systems (cell, internet, social networks), are resilient to component failure

• Network topology plays an important role in this robustness– Even if ~80% of nodes fail, the remaining ~20%

still maintain network connectivity

• Attack vulnerability if hubs are selectively targeted

• In yeast, only ~20% of proteins are lethal when deleted, and are 5 times more likely to have degree k>15 than k<5.

Other interesting features

• Cellular networks are assortative, hubs tend not to interact directly with other hubs.

• Hubs tend to be “older” proteins (so far claimed for protein-protein interaction networks only)

• Hubs also seem to have more evolutionary pressure—their protein sequences are more conserved than average between species (shown in yeast vs. worm)

Sub-cellular localization coverage

Co-localization of interacting proteins

Tendency to interact with your cousin

Over-representation of highly abundant proteins

Coverage versus Accuracy

say a lot, of which most is wrong

say a lot, of which most is right

say little, of which most is wrong

say little, of which most is right

Specificity

Sensitivity